-
公开(公告)号：KR1020100072776A
公开(公告)日：2010-07-01
申请号：KR1020080131284
申请日：2008-12-22
Applicant: 한국전자통신연구원
IPC: G01C11/06
CPC classification number: G01C11/06 , G03B37/00 , G06T7/50 , G06T2207/10028 , H04N19/513
Abstract: PURPOSE: A distance measuring method and a stereo matching system is provided to enable a network mobile robot to easily obtain the distance information based on images without a separate stereo matching hardware or an additional distant sensor. CONSTITUTION: A distance measuring method comprises following steps. A reference motion vector is established by enabling calibration on first and the second cameras. A second camera images is encoded with reference to the first camera image. If blocks of the second camera images is predicted and encoded, the motion vector and the standard motion vector of the blocks are compared. If the compared motion vector is not the same as the standards motion vector, the size value of the motion vector is determined as disparity value.
Abstract translation: 目的：提供距离测量方法和立体匹配系统，使网络移动机器人能够轻松获得基于图像的距离信息，而无需单独的立体匹配硬件或附加远距离传感器。 构成：距离测量方法包括以下步骤。 通过在第一和第二相机上进行校准来建立参考运动矢量。 参考第一相机图像对第二相机图像进行编码。 如果预测和编码第二相机图像的块，则比较块的运动矢量和标准运动矢量。 如果比较运动矢量与标准运动矢量不相同，则将运动矢量的大小值确定为视差值。
-
公开(公告)号：KR1020100019804A
公开(公告)日：2010-02-19
申请号：KR1020080078528
申请日：2008-08-11
Applicant: 한국전자통신연구원
CPC classification number: G06T5/40 , G06T7/593 , G06T2207/10012 , H04N13/133 , H04N13/239 , H04N13/246 , H04N13/00
Abstract: PURPOSE: A stereo vision system and a control method thereof are provided to equalize brightness values of color information of right/left images inputted from right/left stereo cameras and obtain an image according to accurate depth information by compensating right/left images. CONSTITUTION: A stereo vision system comprises the following units. If an image of right/left stereo cameras is inputted, an image information extracting unit(200) extracts color information for adjusting brightness of an image. An image preprocessing unit(300) reduces noise of right/left images using the color information. A stereo matching unit(400) generates depth information by performing stereo matching on the right/left images through an algorithm. A matching result measuring unit(500) receives the depth information to measures a matching degree and changes a pre-stored calibration parameter.
Abstract translation: 目的：提供一种立体视觉系统及其控制方法，以平衡从右/左立体声摄像机输入的左/右图像的颜色信息的亮度值，并通过补偿右/左图像，根据准确的深度信息获得图像。 构成：立体视觉系统包括以下单元。 如果输入右/左立体相机的图像，则图像信息提取单元（200）提取用于调整图像的亮度的颜色信息。 图像预处理单元（300）使用颜色信息降低右/左图像的噪声。 立体匹配单元（400）通过算法对右/左图像执行立体匹配来生成深度信息。 匹配结果测量单元（500）接收深度信息以测量匹配度并改变预先存储的校准参数。
-
公开(公告)号：KR1020080050968A
公开(公告)日：2008-06-10
申请号：KR1020070071863
申请日：2007-07-18
Applicant: 한국전자통신연구원
CPC classification number: B25J19/04
Abstract: An image output device of a network-based intelligent service robot having a stereo camera is provided to calibrate a stereoscopic camera mounted in a robot by having a function of alternately outputting a left pre-processed image and a right pre-processed image one row by one row. The first image selecting unit(1152) selects one of a left camera image and a left pre-processed image. The second image selecting unit(1153) selects one of a right camera image and a right pre-processed image. A left/right image alternately processing unit(1151) alternately outputs the left pre-processed image and the right pre-processed image one row by one row. The third image selecting unit(1154) selects one of the alternately outputted image and a pose-processed image. The first line/frame memory unit(1156) stores and outputs of the first and second image selecting units at 1/4 quadrant and 2/4 quadrant. The second line/frame memory unit(1157) stores and outputs an output of the third image selecting unit and a stereo-matched image to a 3/4 quadrant and a 4/4 quadrant. A resolution matching unit(1155) matches resolution of the outputs of the first to third selecting unit and the stereo matched image, and displays the images of the first and second line/frame memory units as 4-divided screen images.
Abstract translation: 提供具有立体相机的基于网络的智能服务机器人的图像输出装置，通过具有交替地输出左预处理图像和右预处理图像的功能来校准安装在机器人中的立体摄像机 一排 第一图像选择单元（1152）选择左摄像机图像和左预处理图像之一。 第二图像选择单元（1153）选择右侧相机图像和右侧预处理图像之一。 左/右图像交替处理单元（1151）一行一行交替地输出左预处理图像和右预处理图像。 第三图像选择单元（1154）选择交替输出的图像和姿态处理图像中的一个。 第一行/帧存储单元（1156）存储第一和第二图像选择单元的1/4象限和2/4象限的输出。 第二行/帧存储单元（1157）将第三图像选择单元和立体匹配图像的输出存储并输出到3/4象限和4/4象限。 分辨率匹配单元（1155）匹配第一至第三选择单元和立体匹配图像的输出的分辨率，并将第一和第二行/帧存储单元的图像显示为4分屏幕图像。
-
-
公开(公告)号：KR101366776B1
公开(公告)日：2014-02-21
申请号：KR1020090120391
申请日：2009-12-07
Applicant: 한국전자통신연구원 , 세종대학교산학협력단
CPC classification number: G06K9/00241 , G06K9/40
Abstract: 본 발명은 영상 객체를 검출하는 기법에 관한 것으로, 입력 영상의 잡음을 제거하여 저장하고, 저장된 입력 영상의 크기를 조절하며, 크기가 조절된 입력 영상을 MCT 방식으로 변환한 후에, 그 변환 영상에서 MCT값을 이용하여 후보 영역을 검출하고, 검출된 후보 영역들 중에서 사람 영역을 결정하며, 결정된 사람 영역을 표시하여 출력함으로써, 입력 영상에서 영상 객체를 효과적으로 검출할 수 있는 것이다.
생체 인식, 얼굴 인식, 사람 검출-
Patent Agency Ranking
公开(公告)号：KR1020130072073A
公开(公告)日：2013-07-01
申请号：KR1020110139621
申请日：2011-12-21
Applicant: 한국전자통신연구원 , 세종대학교산학협력단
IPC: H04N5/208
CPC classification number: G06K9/4604 , G06F15/16 , G06K9/48
Abstract: PURPOSE: A device and method for an image outline are provided to perform feature point extraction with high reliability even in quantizing noise and complex images by applying a computer vision method and an image signal processing method. CONSTITUTION: A noise removing unit (11) removes noise generated from input images. An inclination calculator (12) calculates the inclination of the image without noise. An inclination intensity calculator (13) calculates the intensity of the inclination. An initial outline detector (14) detects initial outlines from the image without noise using the inclination and intensity in accordance with a first standard., A final outline extractor (15) extracts final outlines among the initial outlines on the basis of a second standard. [Reference numerals] (11) Noise removing unit; (12) Inclination calculator; (13) Inclination intensity calculator; (14) Initial outline detector; (15) Final outline extractor; (16) Power supply unit; (17) Main control unit
Abstract translation: 目的：提供一种用于图像轮廓的装置和方法，以便通过应用计算机视觉方法和图像信号处理方法，即使在量化噪声和复杂图像中也能够以高可靠性执行特征点提取。 构成：噪声去除单元（11）去除输入图像产生的噪声。 倾斜计算器（12）计算没有噪声的图像的倾斜度。 倾斜度计算器（13）计算倾斜度。 初始轮廓检测器（14）使用根据第一标准的倾斜度和强度从无噪声的图像检测初始轮廓。最终轮廓提取器（15）基于第二标准提取初始轮廓中的最终轮廓。 （附图标记）（11）噪声去除单元; （12）倾角计算器; （13）倾角计算器; （14）初始轮廓检测器; （15）最终轮廓提取器; （16）电源单元; （17）主控单元
-
公开(公告)号：KR1020110071213A
公开(公告)日：2011-06-29
申请号：KR1020090127717
申请日：2009-12-21
Applicant: 한국전자통신연구원
IPC: G06T17/00
CPC classification number: G06T15/04 , A63F13/5252 , A63F2300/1087 , A63F2300/695 , G06T7/593 , G06T2200/08 , G06T2207/20028 , G06T2207/30201
Abstract: PURPOSE: A 3D avatar face generation apparatus using stereo vision and a face detector and a method thereof are provided to acquire a desired 3D face avatar according to an abstraction level. CONSTITUTION: A stereo matching unit(103) creates a depth map image from left/right images. A bilateral filter unit(104) abstracts a face image through a bilateral filtering process. A texture mapping unit(107) performs the texture mapping of the abstracted face image to the depth map image. The texture mapping unit creates a 3D avatar.
Abstract translation: 目的：提供使用立体视觉的3D头像面部生成装置和面部检测器及其方法，以根据抽象级别获取期望的3D脸部化身。 构成：立体匹配单元（103）从左/右图像创建深度图图像。 双边过滤器单元（104）通过双边过滤处理抽象脸部图像。 纹理映射单元（107）执行抽象面部图像到深度图图像的纹理映射。 纹理映射单元创建3D头像。
-
公开(公告)号：KR1020100113850A
公开(公告)日：2010-10-22
申请号：KR1020090032372
申请日：2009-04-14
Applicant: 한국전자통신연구원
CPC classification number: G06K9/00255 , G06T7/593 , G06T7/74 , G06T2207/10012 , G06T2207/30201
Abstract: PURPOSE: An apparatus for detecting a face and a method for estimating distance by using the same are provided to implement face detection and distance measurement in a low performance system. CONSTITUTION: A stereo camera unit(100) generates left and right image frames including a face and, outputs the left and right image frames individually. A face detection unit(200) detects a left face block including a face from the left image frame and a right face block from the right image frame. The face detection unit calculates the coordinate difference between the left and right face blocks.
Abstract translation: 目的：提供一种用于检测脸部的装置和使用该装置估计距离的方法，以在低性能系统中实现面部检测和距离测量。 构成：立体相机单元（100）产生包括脸部的左右图像帧，并分别输出左图像帧和右图像帧。 面部检测单元（200）从右图像帧检测包括来自左图像帧的脸部和右脸部区块的左脸部块。 面部检测单元计算左右面块之间的坐标差。
-
公开(公告)号：KR1020100070896A
公开(公告)日：2010-06-28
申请号：KR1020080129632
申请日：2008-12-18
Applicant: 한국전자통신연구원
CPC classification number: G06K9/00228 , G06K9/00973
Abstract: PURPOSE: A face detecting system and a method thereof are provided to detect a face efficiently using a look-up table in real time quickly. CONSTITUTION: A cost calculating unit(240) parallel receives a preprocessing coefficient value corresponding to a window size. The cost calculating unit calculates cost values corresponding to the preprocessing coefficient values. The cost calculating unit adds the calculated cost values. A face pattern discriminating unit(250) compares the added sum of the cost values with a preset threshold value. The face pattern discriminating unit determines whether a preprocessing image corresponding to the window size is a face pattern according to the comparison result.
Abstract translation: 目的：提供一种面部检测系统及其方法，用于快速实时地使用查找表来有效地检测脸部。 构成：成本计算单元（240）并行地接收与窗口尺寸对应的预处理系数值。 成本计算单元计算与预处理系数值对应的成本值。 成本计算单元添加计算出的成本值。 面部图案识别单元（250）将成本值的相加和预先设定的阈值进行比较。 面部图案识别单元根据比较结果确定与窗口尺寸相对应的预处理图像是否是面部图案。
-
公开(公告)号：KR1020100066914A
公开(公告)日：2010-06-18
申请号：KR1020080125425
申请日：2008-12-10
Applicant: 한국전자통신연구원
CPC classification number: G06T7/593
Abstract: PURPOSE: An apparatus and a method for matching an image are provided to apply SAD(Sum of Absolute Difference) and census transform by a stereo matching algorithm and apply a dynamic gamma value, thereby matching images. CONSTITUTION: A decider(510) decides whether a node is a matching-possible region. The first pixel of a left image of a subject and the second pixel of a right image of the subject are calculated in the node. If the node is the matching possible region, a matching region operating unit(520) receives output of the decider. The matching region operating unit calculates SAD and RMCT(Received Mean Census Transform) values for left and right windows.
Abstract translation: 目的：提供一种用于匹配图像的装置和方法以通过立体匹配算法应用SAD（绝对差值之和）和普查变换，并应用动态伽马值，从而匹配图像。 构成：决定器（510）确定节点是否是可匹配的区域。 在节点中计算被摄体的左图像的第一像素和对象的右图像的第二像素。 如果节点是匹配的可能区域，则匹配区域操作单元（520）接收决定器的输出。 匹配区域操作单元计算左窗口和右窗口的SAD和RMCT（接收平均人口普查变换）值。
-
-
-
-
-
-
-
-
-
Search Results
41
records
(took 0.025 seconds)
